Ss2 do intersect efif #9303

efifogel · 2026-01-29T09:20:00Z

Summary of Changes

Enhanced curve-intersection detection:

Made it robust (can be safely used now with a limited precision kernel)
Optimize it for linear curves
Added support for both x-monotone and general curves

Release Management

Affected package(s): Arrangement_on_surface_2, Surface_sweep_2, and Boolean_set_operations_2
Issue(s) solved (if any):
1. Fix Do curves intersect does not work with EPICK Do curves intersect does not work with EPICK #9151
2. Fix the dispatching described in PR Fix SFINAE usage in insert and do_intersect functions on Ubuntu 24.04 #8770
Feature/Small Feature (if any): Intersection Detection Enhancements
Link to compiled documentation:
1. 2D Arrangements
2. 2D Intersection of Curves
License and copyright ownership: TAU

TODO:

move Surface_sweep_2/include/CGAL/variant_output_iterator.h to STL_extension
Check if the new do-intersect should replace the one in the kernel:

You can find a small benchmark program Arrangement_on_surface_2/
benchmark/Arrangement_on_surface_2/do_intersect.cpp that compares the
do_intersect functionality:

The Do_intersect functor of the Arr_non_caching_segment_traits_2

The Do_intersect functor of thekernel

The do_intersect() free function of the kernel
Naturally, the performance of (2.) and (3.) are comparable. However, (1)
is the best.
The code resides under Arrangement_on_surface_2/include/CGAL/
Arrangement_2/do_segments_intersect.h
This code should somehow replace the code in the kernel...

Arrangement_on_surface_2/include/CGAL/Arrangement_2/do_segments_intersect.h

afabri · 2026-02-05T12:13:19Z

Arrangement_on_surface_2/include/CGAL/Arrangement_2/do_segments_intersect.h

+template <typename Point_2, typename Traits>
+bool do_closed_segment_intersect(const Point_2& l1, const Point_2& r1, const Point_2& l2, const Point_2& r2,
+                                 const Traits& traits) {
+  auto cmpare_xy = traits.compare_xy_2_object();


Is the missing letter o on purpose?

Arrangement_on_surface_2/include/CGAL/Arr_circle_segment_traits_2.h

afabri · 2026-02-05T12:20:53Z

Arrangement_on_surface_2/benchmark/Arrangement_on_surface_2/CMakeLists.txt

+cmake_minimum_required(VERSION 3.12...3.31)
+project(do_intersect)
+find_package(Boost CONFIG ${BOOST_MIN_VERSION} REQUIRED COMPONENTS program_options)
+find_package(CGAL REQUIRED COMPONENTS Core OPTIONAL_COMPONENTS Qt6)


Suggested change

find_package(CGAL REQUIRED COMPONENTS Core OPTIONAL_COMPONENTS Qt6)

find_package(CGAL OPTIONAL_COMPONENTS Qt6)

because Core is no longer conditional

afabri · 2026-02-05T12:23:06Z

Surface_sweep_2/include/CGAL/variant_output_iterator.h

+/* A compatible output iterator that accepts a std::variant and dispatches its
+ * alternatives to different sinks via a visitor.
+ *
+ * Designed to replace boost::function_output_iterator when the algorithm emits
+ * heterogeneous output (e.g., `Point_2` or `X_monotone_curve_2`).
+ *
+ * Many CGAL concepts (e.g., Make_x_monotone_2) require an operation that writes
+ * results to an `OutputIterator` whose value type is:
+ *
+ * `std::variant<T1, T2, ...>`
+ *
+ * However:
+ * 1. `boost::function_output_iterator` receives one type
+ * 2. Visitors (operator()) are invoked after assignment
+ * 3. CGAL freely copies, assigns, and default-constructs output iterators
+ * `variant_output_iterator` bridges this gap.


Suggested change

/* A compatible output iterator that accepts a std::variant and dispatches its

* alternatives to different sinks via a visitor.

*

* Designed to replace boost::function_output_iterator when the algorithm emits

* heterogeneous output (e.g., `Point_2` or `X_monotone_curve_2`).

*

* Many CGAL concepts (e.g., Make_x_monotone_2) require an operation that writes

* results to an `OutputIterator` whose value type is:

*

* `std::variant<T1, T2, ...>`

*

* However:

* 1. `boost::function_output_iterator` receives one type

* 2. Visitors (operator()) are invoked after assignment

* 3. CGAL freely copies, assigns, and default-constructs output iterators

* `variant_output_iterator` bridges this gap.

/* A compatible output iterator that accepts a `std::variant` and dispatches its

* alternatives to different sinks via a visitor.

*

* Designed to replace `boost::function_output_iterator` when the algorithm emits

* heterogeneous output (e.g., `Point_2` or `X_monotone_curve_2`).

*

* Many \cgal concepts (e.g., `Make_x_monotone_2`) require an operation that writes

* results to an `OutputIterator` whose value type is:

*

* `std::variant<T1, T2, ...>`

*

* However:

* 1. `boost::function_output_iterator` receives one type

* 2. Visitors (operator()) are invoked after assignment

* 3. \cgal freely copies, assigns, and default-constructs output iterators

* `variant_output_iterator` bridges this gap.

Make_x_monotone_2 is not a concept. CamelCase ?

…ts", which reflects the true semantics of the parameter

… Ss2-do_intersect-efif

efifogel · 2026-02-05T13:44:45Z

Yes it is; fixed. ____ _ ____ _ /_____/_) o /__________ __ // (____ ( ( ( (_/ (_/-(-'_(/ _/

…

On Thu, 5 Feb 2026 at 14:24, Andreas Fabri ***@***.***> wrote: ***@***.**** commented on this pull request. ------------------------------ In Surface_sweep_2/include/CGAL/variant_output_iterator.h <#9303 (comment)>: > +/* A compatible output iterator that accepts a std::variant and dispatches its + * alternatives to different sinks via a visitor. + * + * Designed to replace boost::function_output_iterator when the algorithm emits + * heterogeneous output (e.g., `Point_2` or `X_monotone_curve_2`). + * + * Many CGAL concepts (e.g., Make_x_monotone_2) require an operation that writes + * results to an `OutputIterator` whose value type is: + * + * `std::variant<T1, T2, ...>` + * + * However: + * 1. `boost::function_output_iterator` receives one type + * 2. Visitors (operator()) are invoked after assignment + * 3. CGAL freely copies, assigns, and default-constructs output iterators + * `variant_output_iterator` bridges this gap. Make_x_monotone_2 is not a concept. CamelCase ? — Reply to this email directly, view it on GitHub <#9303 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/ABVBNOD6VA4OTVAK7J4FVXD4KMY6LAVCNFSM6AAAAACTJBWSXKVHI2DSMVQWIX3LMV43YUDVNRWFEZLROVSXG5CSMV3GSZLXHMZTONJWG44TINJQHE> . You are receiving this because you authored the thread.Message ID: ***@***.***>

efifogel · 2026-02-05T13:50:49Z

Added the missing *backticks* ____ _ ____ _ /_____/_) o /__________ __ // (____ ( ( ( (_/ (_/-(-'_(/ _/

…

On Thu, 5 Feb 2026 at 14:23, Andreas Fabri ***@***.***> wrote: ***@***.**** commented on this pull request. ------------------------------ In Surface_sweep_2/include/CGAL/variant_output_iterator.h <#9303 (comment)>: > +/* A compatible output iterator that accepts a std::variant and dispatches its + * alternatives to different sinks via a visitor. + * + * Designed to replace boost::function_output_iterator when the algorithm emits + * heterogeneous output (e.g., `Point_2` or `X_monotone_curve_2`). + * + * Many CGAL concepts (e.g., Make_x_monotone_2) require an operation that writes + * results to an `OutputIterator` whose value type is: + * + * `std::variant<T1, T2, ...>` + * + * However: + * 1. `boost::function_output_iterator` receives one type + * 2. Visitors (operator()) are invoked after assignment + * 3. CGAL freely copies, assigns, and default-constructs output iterators + * `variant_output_iterator` bridges this gap. ⬇️ Suggested change -/* A compatible output iterator that accepts a std::variant and dispatches its - * alternatives to different sinks via a visitor. - * - * Designed to replace boost::function_output_iterator when the algorithm emits - * heterogeneous output (e.g., `Point_2` or `X_monotone_curve_2`). - * - * Many CGAL concepts (e.g., Make_x_monotone_2) require an operation that writes - * results to an `OutputIterator` whose value type is: - * - * `std::variant<T1, T2, ...>` - * - * However: - * 1. `boost::function_output_iterator` receives one type - * 2. Visitors (operator()) are invoked after assignment - * 3. CGAL freely copies, assigns, and default-constructs output iterators - * `variant_output_iterator` bridges this gap. +/* A compatible output iterator that accepts a `std::variant` and dispatches its + * alternatives to different sinks via a visitor. + * + * Designed to replace `boost::function_output_iterator` when the algorithm emits + * heterogeneous output (e.g., `Point_2` or `X_monotone_curve_2`). + * + * Many \cgal concepts (e.g., `Make_x_monotone_2`) require an operation that writes + * results to an `OutputIterator` whose value type is: + * + * `std::variant<T1, T2, ...>` + * + * However: + * 1. `boost::function_output_iterator` receives one type + * 2. Visitors (operator()) are invoked after assignment + * 3. \cgal freely copies, assigns, and default-constructs output iterators + * `variant_output_iterator` bridges this gap. — Reply to this email directly, view it on GitHub <#9303 (review)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/ABVBNOD4BHFP54DRYYBFHGT4KMY4DAVCNFSM6AAAAACTJBWSXKVHI2DSMVQWIX3LMV43YUDVNRWFEZLROVSXG5CSMV3GSZLXHMZTONJWG44TEMJXG4> . You are receiving this because you authored the thread.Message ID: ***@***.***>

efifogel · 2026-02-05T13:53:32Z

done ____ _ ____ _ /_____/_) o /__________ __ // (____ ( ( ( (_/ (_/-(-'_(/ _/

…

On Thu, 5 Feb 2026 at 14:21, Andreas Fabri ***@***.***> wrote: ***@***.**** commented on this pull request. ------------------------------ In Arrangement_on_surface_2/benchmark/Arrangement_on_surface_2/CMakeLists.txt <#9303 (comment)>: > @@ -0,0 +1,6 @@ +cmake_minimum_required(VERSION 3.12...3.31) +project(do_intersect) +find_package(Boost CONFIG ${BOOST_MIN_VERSION} REQUIRED COMPONENTS program_options) +find_package(CGAL REQUIRED COMPONENTS Core OPTIONAL_COMPONENTS Qt6) ⬇️ Suggested change -find_package(CGAL REQUIRED COMPONENTS Core OPTIONAL_COMPONENTS Qt6) +find_package(CGAL OPTIONAL_COMPONENTS Qt6) because Core is no longer conditional — Reply to this email directly, view it on GitHub <#9303 (review)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/ABVBNOHLBPDUZCOSGMB3P3L4KMYT3AVCNFSM6AAAAACTJBWSXKVHI2DSMVQWIX3LMV43YUDVNRWFEZLROVSXG5CSMV3GSZLXHMZTONJWG44DGMZRGI> . You are receiving this because you authored the thread.Message ID: ***@***.***>

efifogel · 2026-02-05T13:57:40Z

No, I fixed it ____ _ ____ _ /_____/_) o /__________ __ // (____ ( ( ( (_/ (_/-(-'_(/ _/

…

On Thu, 5 Feb 2026 at 14:13, Andreas Fabri ***@***.***> wrote: ***@***.**** commented on this pull request. ------------------------------ In Arrangement_on_surface_2/include/CGAL/Arrangement_2/do_segments_intersect.h <#9303 (comment)>: > + +#include <CGAL/license/Arrangement_on_surface_2.h> + +/*! \file Intersection detection functions. + */ + +namespace CGAL { +namespace Aos_2 { +namespace internal { + +// Specialized do_intersect with several tests skipped because at +// this point, we already know the order of points +template <typename Point_2, typename Traits> +bool do_closed_segment_intersect(const Point_2& l1, const Point_2& r1, const Point_2& l2, const Point_2& r2, + const Traits& traits) { + auto cmpare_xy = traits.compare_xy_2_object(); Is the missing letter o on purpose? — Reply to this email directly, view it on GitHub <#9303 (review)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/ABVBNODHNFHPXGVEA3SX3SD4KMXXNAVCNFSM6AAAAACTJBWSXKVHI2DSMVQWIX3LMV43YUDVNRWFEZLROVSXG5CSMV3GSZLXHMZTONJWG42DCMJYGM> . You are receiving this because you authored the thread.Message ID: ***@***.***>

sloriot · 2026-02-10T07:55:21Z

Several runtime errors in CGAL-6.2-Ic-106

afabri · 2026-02-10T07:59:37Z

https://cgal.geometryfactory.com/CGAL/testsuite/summary-6.2-Ic-106.html?package=Arrangement_on_surface_2

…loading (some compilers complaine).

efifogel · 2026-02-10T08:29:29Z

I fixed them ____ _ ____ _ /_____/_) o /__________ __ // (____ ( ( ( (_/ (_/-(-'_(/ _/

…

On Tue, 10 Feb 2026 at 09:58, Sebastien Loriot ***@***.***> wrote: *sloriot* left a comment (CGAL/cgal#9303) <#9303 (comment)> Several runtime errors in CGAL-6.2-Ic-106 — Reply to this email directly, view it on GitHub <#9303 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/ABVBNOBNASVVAHIFLKFMIED4LGFH7AVCNFSM6AAAAACTJBWSXKVHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMZTQNZVHE3TAMJRG4> . You are receiving this because you authored the thread.Message ID: ***@***.***>

efifogel · 2026-02-10T08:58:37Z

I also fixed the conflict in CHANGES.md (and fixed the content of my specific change...) ____ _ ____ _ /_____/_) o /__________ __ // (____ ( ( ( (_/ (_/-(-'_(/ _/

…

On Tue, 10 Feb 2026 at 10:29, Efi Fogel ***@***.***> wrote: I fixed them ____ _ ____ _ /_____/_) o /__________ __ // (____ ( ( ( (_/ (_/-(-'_(/ _/ On Tue, 10 Feb 2026 at 09:58, Sebastien Loriot ***@***.***> wrote: > *sloriot* left a comment (CGAL/cgal#9303) > <#9303 (comment)> > > Several runtime errors in CGAL-6.2-Ic-106 > > — > Reply to this email directly, view it on GitHub > <#9303 (comment)>, or > unsubscribe > <https://github.com/notifications/unsubscribe-auth/ABVBNOBNASVVAHIFLKFMIED4LGFH7AVCNFSM6AAAAACTJBWSXKVHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMZTQNZVHE3TAMJRG4> > . > You are receiving this because you authored the thread.Message ID: > ***@***.***> >

albert-github · 2026-02-10T09:03:28Z

Arrangement_on_surface_2/doc/Arrangement_on_surface_2/Concepts/AosTraits--DoIntersect_2.h

+   * \param xc2 The second \f$x\f$-monotone curve.
+   * \param consider_common_endpoints indicates whether common endpoints should be counted as intersections.
+   * \return `true` if `consider_common_endpoints` is true and `xcv1` and `xcv2` intersect or if
+   *  `consider_common_endpoints` is `false and at least one of the interiors of `xcv1` and `xcv2` intersect,


Suggested change

* `consider_common_endpoints` is `false and at least one of the interiors of `xcv1` and `xcv2` intersect,

* `consider_common_endpoints` is `false` and at least one of the interiors of `xcv1` and `xcv2` intersect,

efifogel · 2026-02-10T10:30:23Z

Fixed ____ _ ____ _ /_____/_) o /__________ __ // (____ ( ( ( (_/ (_/-(-'_(/ _/

…

On Tue, 10 Feb 2026 at 11:04, albert-github ***@***.***> wrote: ***@***.**** commented on this pull request. ------------------------------ In Arrangement_on_surface_2/doc/Arrangement_on_surface_2/Concepts/AosTraits--DoIntersect_2.h <#9303 (comment)>: > + * \cgalHasModelsEnd + */ +class DoIntersect_2 { +public: + + /// \name Operations + /// A model of this concept must provide: + /// @{ + + /*! determines whether two \f$x\f$-monotone curves intersect. + * + * \param xc1 The first \f$x\f$-monotone curve. + * \param xc2 The second \f$x\f$-monotone curve. + * \param consider_common_endpoints indicates whether common endpoints should be counted as intersections. + * \return `true` if `consider_common_endpoints` is true and `xcv1` and `xcv2` intersect or if + * `consider_common_endpoints` is `false and at least one of the interiors of `xcv1` and `xcv2` intersect, ⬇️ Suggested change - * `consider_common_endpoints` is `false and at least one of the interiors of `xcv1` and `xcv2` intersect, + * `consider_common_endpoints` is `false` and at least one of the interiors of `xcv1` and `xcv2` intersect, — Reply to this email directly, view it on GitHub <#9303 (review)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/ABVBNOHWDXZYI5MG5KNNQS34LGNIBAVCNFSM6AAAAACTJBWSXKVHI2DSMVQWIX3LMV43YUDVNRWFEZLROVSXG5CSMV3GSZLXHMZTONZXGY2DKNZYGM> . You are receiving this because you authored the thread.Message ID: ***@***.***>

efifogel added 2 commits January 29, 2026 19:14

Introduced Do_intersect_surface_sweep_2

94d0e5b

Clean implementation of Surface_sweep_2::do_intersect

ae33c8d

sloriot force-pushed the Ss2-do_intersect-efif branch from b277e45 to ae33c8d Compare January 29, 2026 18:14

sloriot added 4 commits January 29, 2026 19:22

add missing license

5bdda0a

missing license

98a924c

trailing whitespaces

25373ec

remove tmp files

dc64fdb

sloriot added Enhancement Not yet approved The feature or pull-request has not yet been approved. Pkg::Arrangement_on_surface_2 TODO labels Jan 29, 2026

afabri reviewed Feb 5, 2026

View reviewed changes

Arrangement_on_surface_2/include/CGAL/Arrangement_2/do_segments_intersect.h Outdated Show resolved Hide resolved

afabri and others added 3 commits February 5, 2026 11:12

Apply suggestion from @afabri

670d6d9

Added missing return type

8615bcb

Replaced "closed" with "consider_end_points"

a328685

afabri reviewed Feb 5, 2026

View reviewed changes

Arrangement_on_surface_2/include/CGAL/Arr_circle_segment_traits_2.h Outdated Show resolved Hide resolved

Apply suggestion from @afabri

ba038c6

afabri reviewed Feb 5, 2026

View reviewed changes

efifogel added 3 commits February 5, 2026 15:33

Changed names of parameters from "closed" => "consider_common_endpoin…

f22869a

…ts", which reflects the true semantics of the parameter

Merge branch 'Ss2-do_intersect-efif' of github.com:efifogel/cgal into…

d74aa6d

… Ss2-do_intersect-efif

Fixed name of concept (Make_x_monotone_2 => AosTraits::MakeXMonotone_2)

92d79bd

Added missing backticks

b487bb3

Removed requirement of "Core"

541fd11

Fixed typo in name of variable

79897d0

sloriot added 2 commits February 9, 2026 15:37

fix warnings

150b457

fix compilation error

da31850

sloriot added the Under Testing label Feb 9, 2026

sloriot added Tests failing and removed Under Testing labels Feb 10, 2026

efifogel added 2 commits February 10, 2026 10:13

Removed unused variable

9e15f5a

Removed redundant do_intersect() functions and renamed to reduce over…

b0ee78a

…loading (some compilers complaine).

Merge branch 'main' into Ss2-do_intersect-efif

5959247

albert-github reviewed Feb 10, 2026

View reviewed changes

Fixed backticks of true and false

ab49394

sloriot added 2 commits February 10, 2026 14:52

fix warning

c396a7e

add missing include directions

01b3362

	find_package(CGAL REQUIRED COMPONENTS Core OPTIONAL_COMPONENTS Qt6)
	find_package(CGAL OPTIONAL_COMPONENTS Qt6)

	* `consider_common_endpoints` is `false and at least one of the interiors of `xcv1` and `xcv2` intersect,
	* `consider_common_endpoints` is `false` and at least one of the interiors of `xcv1` and `xcv2` intersect,

Ss2 do intersect efif #9303

Are you sure you want to change the base?

Ss2 do intersect efif #9303

Uh oh!

Conversation

efifogel commented Jan 29, 2026 • edited by sloriot Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Summary of Changes

Release Management

TODO:

Uh oh!

Uh oh!

afabri Feb 5, 2026

Choose a reason for hiding this comment

Uh oh!

Uh oh!

afabri Feb 5, 2026

Choose a reason for hiding this comment

Uh oh!

afabri Feb 5, 2026

Choose a reason for hiding this comment

Uh oh!

afabri Feb 5, 2026

Choose a reason for hiding this comment

Uh oh!

efifogel commented Feb 5, 2026 via email

Uh oh!

efifogel commented Feb 5, 2026 via email

Uh oh!

efifogel commented Feb 5, 2026 via email

Uh oh!

efifogel commented Feb 5, 2026 via email

Uh oh!

sloriot commented Feb 10, 2026

Uh oh!

afabri commented Feb 10, 2026

Uh oh!

efifogel commented Feb 10, 2026 via email

Uh oh!

efifogel commented Feb 10, 2026 via email

Uh oh!

albert-github Feb 10, 2026

Choose a reason for hiding this comment

Uh oh!

efifogel commented Feb 10, 2026 via email

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

4 participants

efifogel commented Jan 29, 2026 •

edited by sloriot

Loading